Please use this identifier to cite or link to this item: http://dx.doi.org/10.18419/opus-10467
Authors: Zeller, Marcel
Title: Automated detection and extraction of deployment components of Docker services
Issue Date: 2019
metadata.ubs.publikation.typ: Abschlussarbeit (Master)
metadata.ubs.publikation.seiten: 88
URI: http://nbn-resolving.de/urn:nbn:de:bsz:93-opus-ds-104841
http://elib.uni-stuttgart.de/handle/11682/10484
http://dx.doi.org/10.18419/opus-10467
Abstract: Validation of deployability of application topologies at design-time is a difficult task. Often application topologies cannot be checked until the complete system is deployed. A reliable and up-to-date data source that could be used for validation is a big problem. A possible solution can be the use of the build-scripts of Docker images: Dockerfiles. They provide application component relationships in a well-defined syntax. To be able to extract reliable software component relationship data from dockerfiles a big source set is necessary. Online software repository services provide many public available dockerfiles. Crawlers can be used to collect them. Afterwards the dockerfiles need to be analyzed. Ambiguous, misleading or incorrect content of dockerfiles must be taken into account as far as possible to ensure reliable results. This thesis outlines a concept for collection of dockerfiles, discovery and extraction of deployment components and understandable depiction of the results. The collection is done with crawlers for specific online services. For the discovery and extraction analyzer for component installation commands are provided. A metamodel for models of component structures ensures the comprehensibility of the results. The concept is implemented prototypically with the support of one online software repository service and several wide-spread installation command-line tools.
Appears in Collections:05 Fakultät Informatik, Elektrotechnik und Informationstechnik

Files in This Item:
File Description SizeFormat 
ThesisMarcelZeller.pdf1,81 MBAdobe PDFView/Open


Items in OPUS are protected by copyright, with all rights reserved, unless otherwise indicated.